BREAKING CHANGES
The biggest breaking change is that now all the functions always return an expression containing log value of Bayes Factor in favor of the null hypothesis (\(log(BF_{01})\)). This simplifies the code and also reduces confusion. The evidence in favor of alternative is of course just negative of this value.
Removes the unnecessary aliases for certain tests: bf_one_sample_ttest, bf_two_sample_ttest, and bf_onesample_proptest.
The output dataframe now only contains results for the alternative hypothesis (BF10) and its log value to avoid cluttered dataframe since all the other columns were essentially mathematical transformations of it and not really relevant.
The output argument’s default value has been changed from too generic "results" to more specific and informative "dataframe".
The argument caption has been renamed to top.text since caption doesn’t really have much sense outside of the plotting-context of ggstatsplot in which it was originally conceived.
The t-test and ANOVA tests get subject.id argument relevant for repeated measures design.
Renames bf_meta to bf_meta_random. The previous alias is retained for now but will be retired in future releases.
MAJOR CHANGES
Replaces the heavier broomExtra dependency with parameters.
bf_oneway_anova now returns model-averaged R-squared posterior as a measure of effect size.
BREAKING CHANGES
bf_expr now expects not a dataframe, but rather the BayesFactor object.MAJOR CHANGES
bf_extractor now internally uses parameters::model_parameters to extract as many details from the BayesFactor as possible. So the returned dataframe will contain additional details now.
The redundant text related to whether the results in favor of the null or the alternative hypothesis has now been removed, since this information is already present in BF subscripts (01: null, 10: alternative).
MAJOR CHANGES
ggstatsplot-verse, the minimum R version is bumped to 3.6.0.MINOR CHANGES
bf_contingency_tab function no longer includes sampling plan information in the expression.
Internal refactoring to improve data wrangling using ipmisc.
MAJOR CHANGES
bf_meta deprecates messages argument. If users want to see the full output from metaBMA, they are advised to run it independently.
All relevant functions gain a hypothesis.text argument to remove the redundant pretext before the expression containing the results.
MINOR CHANGES
Thanks to Sarah, the package has a hexsticker. :)
Re-exports metaBMA::prior() for statsExpressions.
Removed unnecessary dependence on broomExtra.